We use a patched version of Heritrix. The patch fixes the following issues:
ARC records of >2GB caused arithmetic overflow and could not be read.
- ...something about skipping long records...need to check it...
Patch against Heritrix 1.12.1
Patched sourcecode Heritrix-1.12.1b-src.zip
Patched binary Heritrix-1.12.1b.zip